Mobile Computing in Digital Ecosystems: Design Issues and...

Gabriele D’Angelo<[email protected]>

http://www.cs.unibo.it/gdangelo/

joint work with:Stefano Ferretti, Vittorio Ghini, Fabio Panzieri

Istanbul, Turkey

International Wireless Communications and Mobile Computing conference

(IWCMC), 2011

Mobile Computing in

Digital Ecosystems:

Design Issues and Challenges

2International Wireless Communications and Mobile Computing conference (IWCMC), 2011 Gabriele D'Angelo

Presentation outline

Problem definition and proposal description

Inside the Digital Organism

The Digital Ecosystem

Proposed approach: desiderata and feasibility

Optimizing the Digital Organism

Digital Organisms and the Digital Ecosystem

Seamless connectivity

Optimizing the Digital Ecosystem

Data dissemination in the Digital Ecosystem: gossip based

Computation and storage issues

Conclusions and future work


Problem definition: many devices


Problem definition: with different characteristics

CPUbatteryGSM/UMTS/...WLAN/BluetoothGPSstoragesensors



CPUbatteryGPSstorageBluetooth



sensorsradio protocolbattery



CPUbattery

WLAN/BluetoothGPS

storage



CPUbattery

GSM/UMTS/...WLAN

BluetoothGPS

storagesensors



CPUbattery

GSM/UMTS/...WLAN

storage


Problem definition: with different characteristicsCPU

batterystorage

Bluetooth


Problem definition: a Digital Organism (DO)

CPUbattery

GSM/UMTS/...WLAN

storage

CPUbattery

GSM/UMTS/...WLAN

BluetoothGPS

storagesensors



Problem definition: a Digital Organism (DO)

CPUbattery

GSM/UMTS/...WLAN

storage

CPUbattery

GSM/UMTS/...WLAN

BluetoothGPS

storagesensors


CPUbattery

WLAN/BluetoothGPS

storage


Problem definition: an extended Digital Organism (eDO)



CPUBluetoothGPSmaps


sensors



CPUBluetooth

storage



battery!!!



external antenna





Problem definition: a Digital Ecosystem (DE)


Inside the Digital Organism

A Digital Organism (DO) is composed of many different

devices, each with a specific technical profile

Some characteristics are static:

hardware

communication and computation capabilities

Others are dynamic: for example, the battery level

There are many technological limitations to overcome, each

device is seen as separate tool

The human user often has to do complex configurations on

each device and take decisions on many low-level details


Inside the Digital Organism: a few examples

Short range wireless technologies (e.g. Bluetooth, ZigBee)

can be used for the intra-DO communications (often more

efficient than WLAN or UMTS)

The DO should be an integrated community of devices

One (or more) devices have to be used as “gateways” for

inter-DO communications (e.g. to reach Internet)

How is the gateway device chosen? The “best available

network” should be used

What is “best”? Is it the one with the lowest latency? Higher

bandwidth? Cheapest? More energy efficient?

Many factors... such as the battery level of each device, the

telco rates and policies


The Digital Ecosystem

The Digital Ecosystem (DE) is populated by DOs

DOs can enter and exit the DE: very dynamic behavior

Each DO is a computation and communication resource

It can forward messages (e.g. multi-hop network) but this is

not its main role

Smart P2P schemes must be used among DOs to share

data and resources (e.g. peripherals)

How is the DE organized? What is its topology? How

resources can be accessed by other DOs? What access

control schemes have to be used?


Proposed approach: 4 main points

All the devices within a DO must be integrated using auto-

configuration strategies

Each DO is equipped with a software module capable of multi-

criteria adaptive decision schemes

Smart P2P schemes to share data/resources among DOs:

social aspects (e.g. friends, colleagues, family)

trust, security and privacy

smart discovery of resources

strategies for resources access (e.g. tit-for-tat schemes)

Creation of a global overlay network with some desired

topology, to guarantee an effective data dissemination


Desiderata and feasibility

Novel interaction algorithms: “optimal” organization of

resources both at global and local level

Goal: building an ubiquitous architecture in which the devices

are configured transparently, dynamically and adaptively

depending on:

the characteristics of the device

the environment in which it is deployed

the other users, their devices and the interactions

Most of the technologies for these target paradigms are

already available

What is missing is their integration!


Optimizing the digital organism

Full interaction among all computing resources is needed

It requires to optimize the use of networks that are

available to the user's devices

Decisional procedures based on several criteria

The goal is to model each DO as a computational

environment and to find the best configuration for it

All nodes must broadcast (to the whole DO) their technical

details (and working state) to enable a proper configuration

of other devices in the DOs

A coordination-based approach is quite promising:

distributed algorithm for its election have to be implemented


Digital organism and digital ecosystem

Each DO needs a set of protocols to interact with peer DOs

in the same home ecosystem

What is the best available network? Multi-criteria choice:

bandwidth

connection cost

battery state and consumption

probability to maintaining the connection active while

moving (to minimize hand-offs)

And furthermore... enable the simultaneous and

opportunistic use of all the available networks


Digital organism and digital ecosystem

Following a bottom-up approach, we have started investigating

in deep some specific aspects

Goal: allow the use of multiple network interfaces

transparent at the application level

Proposed implementation: cross-layer architecture that

uses a proxy-based system to offer continuity in the

communication of a given mobile node with a remote proxy

Testbed: a mobile node was equipped with two network

interfaces (Wi-Fi and UMTS)

the coverage of both technologies depends on the node position

is it possible to have a seamless switch without interrupting the

end-to-end communications?


Seamless connectivity: a testbed


Optimizing the digital ecosystem

The DOs have to be organized as a set of interacting nodes

Many approaches are possible (e.g. overlay or mesh networks)

The theory of complex network can help identify the more

adequate topology (e.g. random, scale-free, small-word)

Other factors have to be considered, such as local strategies

based on nodes targets, resources, services and the

human identities behind the DOs (privacy concerns)

(Semi)automatic configuration mechanisms based on:

nodes topology in the environment

type of network to be created

type of services

node characteristics in terms of computation / communication

clustering / proximity (also in terms of social features)


Data dissemination in the digital ecosystem

Dissemination strategies must be adopted to broadcast

messages (e.g. for the discovery of DOs, resources, paths for

messages)

The network is highly dynamic: static and centralized

approaches are not feasible

Unstructured P2P solutions are much more promising: for

example gossip dissemination strategies

Many advantages: fully decentralized, based on local

information, very robust and simple

Drawbacks: high overhead (in some conditions)

Using simulation, we have investigated the most common

gossip protocols (conditional broadcast, fixed probability,

fixed fanout) and proposed a new generation of adaptive

protocols


Evaluation: gossip protocols








Computation and storage issues

Computation demanding queries can be handled in an

adaptive manner: a cloud based system over mobile ad-

hoc technologies

The computations can be performed on unknown hosts,

trusted servers or hosts in the same user domain

The best allocation strategy has to be found, but properties as

privacy, accountability and non reputability must be

guaranteed

Something similar also for remote storage: the best storage

for some data is outside the local DO, somewhere in the DE or

even on Internet


Conclusions and future work

We have discussed a methodology to optimize interactions of

mobile users

Each user is seen as a Digital Organism (DO) that lives in dynamic

and heterogeneous ecosystem, called Digital Environment (DE)

The idea is to optimize the use and interaction of devices inside

each DO and the interactions between DOs in the DE

The technologies needed to build such a system are mostly

already available, what is missing is their integration

Following a bottom-up approach we have investigated some

basic aspects such as seamless connectivity and gossip

dissemination schemes

The next step is the detailed design and implementation of the

rest of the system


Further information

Gabriele D'Angelo, Stefano Ferretti, Vittorio Ghini, Fabio Panzieri

Mobile Computing in Digital Ecosystems: Design Issues and Challenges

Proceedings of the 7th International wireless Communications and Mobile

Computing conference (IWCMC 2011). Istanbul, Turkey, July 2011

A draft version of this paper is freely available at the following link:

http://arxiv.org/pdf/1105.2458v1

Gabriele D'Angelo

E-mail: <[email protected]>


http://arxiv.org/pdf/1105.2458v1

mailto:[email protected]


Gabriele D’Angelo<[email protected]>


joint work with:Stefano Ferretti, Vittorio Ghini, Fabio Panzieri

Istanbul, Turkey

International Wireless Communications and Mobile Computing conference

(IWCMC), 2011

Mobile Computing in

Digital Ecosystems:

Design Issues and Challenges

Date post:	05-Oct-2020
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

Mobile Computing in Digital Ecosystems: Design Issues and...

Documents